Online Learning Probabilistic Event Calculus Theories in Answer Set Programming

Abstract Complex Event Recognition (CER) systems detect event occurrences in streaming time-stamped input using predefined patterns. Logic-based approaches are of special interest CER, since, via Statistical Relational AI, they combine uncertainty-resilient reasoning with time and change, machine learning, thus alleviating the cost manual pattern authoring. We present a system based on Answer Set Programming (ASP), capable probabilistic complex patterns form weighted rules Calculus, whose structure weights learnt online. compare our ASP-based implementation Markov one number state-of-the-art batch learning algorithms CER data sets for activity recognition, maritime surveillance fleet management. Our results demonstrate superiority novel approach, both terms efficiency predictive performance. This paper is under consideration publication Theory Practice Logic (TPLP).